PK10559: ABEND ASOC IN DFHSOM02 BROUGHT DOWN REGION

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Customer is receiving ASOC abend. After taking an SVC dump, two
  additional abends followed within 3 minutes.
  The final abend, an X`0377` in DFHRMLSO resulted in the region
  abending U1800.
  The trace shows
  56337 QR  LMLM  ENTRY TEST_LOCK_OWNER       2C0044E8
  56337 QR  LMLM  EXIT  TEST_LOCK_OWNER/EXCEPTION NOT_LOCK_OWNER
  56337 QR  LMLM  ENTRY LOCK                  2C0044E8,EXCLUSIVE
  56337 QR  LMLM  EXIT  LOCK/EXCEPTION        LOCK_TOKEN_NOT_FOUND
  56337 QR  LMLM  ENTRY DELETE_LOCK           2C0044E8
  56337 QR  LMLM  EXIT  DELETE_LOCK/EXCEPTION NOT_LOCK_OWNER
  56337 QR  SOLT  *EXC* DELETE_LOCK_FAILURE   01226337,0,0,56337
  .
  The kernel shows
  0067 1B562B00 ***Running**1A9B6680 56337 CWXN 27188580
  1A398020 *YES*
  .
  The stacks shows
  KE_N @STACK   TYPE ADDRESS  LINK REG OFFS ER NAME
   0067 1B578020 Bot  9A201600 9A2018AC 02AC    DFHKETA
   0067 1B578150 Dom  9A216420 9A216526 0106    DFHDSKE
   0067 1B5783B0 Dom  9A237148 9A238286 113E    DFHXMTA
   0067 1B5788D0 Dom  9A2B1D38 9A2B2C02 0ECA    DFHRMUO
   0067 1B578E70 Sub  9A290E60 9A291180 0320    DFHRMLSO
   0067 1B579520 Dom  9AC00000 9AC03D72 3D72    DFHSOCK
                 Int     +4530 9AC00240 0240
  PERFORM_COMMIT_SOCK
                 Int     +1F98 9AC046E4 46E4    CLOSE_SOCK
                 Int     +3D24 9AC02090 2090    REMOVE_STE
   0067 1B57A350 Sub  9AC19CE8 9AC1A65E 0976 *Y DFHSOM02
                 Int     +08A0 9AC1A366 067E
  FUNCTIONAL_RECOVERY
   0067 1B57A6C0 Dom  9A276010 9A279996 3986    DFHMEME
                 Int     +2EFC 9A276192 0182    SEND
                 Int     +153C 9A278FFA 2FEA    CONTINUE_SEND
                 Int     +38AE 9A2775E6 15D6
  TAKE_A_DUMP_FOR_CALLER
   0067 1B57B650 Dom  9A303C18 9A30500E 13F6    DFHDUDU
                 Int     +0A1E 9A303D10 00F8    SYSTEM_DUMP
                 Int     +1826 9A304A64 0E4C    TAKE_SYSTEM_DUMP
  .
  The basic problem occurs because CSOL deleted a socket object
  while task 56337 was still using it (mid-syncpoint).
  Additional Symptoms and Keywords:
  ABENDAWSC
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED: All.                                         *
  ****************************************************************
  * PROBLEM DESCRIPTION: ABENDASOC followed by ABEND0C4 in       *
  *                      DFHSOCK CLOSE_SOCK during end of task   *
  *                      syncpoint causing CICS to abend.        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  A web receiver task enters end-of-task syncpoint processing
  following an earlier receive error. DFHSOCK is called for
  function PERFORM_COMMIT. The socket is closed but the CWXN task
  then loses control attempting to acquire the SOLOCK. The SOLOCK
  is owned by the sockets listener task which is performing an
  IMMCLOSE of the TCPIPSERVICE against which the CWXN task was
  attached. IMMCLOSE processing loops through all the client
  sockets associated with the TCPIPSERVICE and cleans them up
  providing no socket sends or receives are currently in progress.
  This cleanup causes the CWXN socket to be freed and its locks
  deleted. The associated STE control block is also freemained.
  This process is all done while holding the global SOLOCK.
  When the sockets listener task has finished cleanup, it releases
  the SOLOCK and the CWXN task now regains control. CWXN tries to
  remove the freemained STE.
  If the STE storage has been re-used and overwritten, DFHSOCK
  may attempt to delete the already deleted socket. This will
  lead to an ABENDASOC when an attempt is made to delete the
  socket's SOLTLOCK.
  A reuse of the STE storage can also lead to a program check
  in DFHSOCK when it attempts to remove the STE from the
  TCPIPSERVICEs STE chain. If this happens, a disaster response
  will be returned to recovery manager which will then issue
  message DFHRM0002 error code x'0377' and CICS is abended.
  .
  If the STE has not been reused, a storage violation will occur
  when DFHSOCK attempts to freemain the already freemained STE.
  
  ADDITIONAL KEYWORDS :- MSGDFHRM0002 RM0002 DFHSM0002 SM0002
                         MSGDFHSM0002 0305 ASOC 0377 0C4 S0C4
                         ABEND0C4 ABENDS0C4
  

Problem conclusion

• DFHSOCK routine close_sock has been changed to verify that the
  socket being closed is still valid before proceeding to delete
  it and its associated STE.
  

Temporary fix

•             *********
              * HIPER *
              *********
  FIX AVAILABLE BY PTF ONLY
  

Comments

• ž**** PE05/11/24 FIX IN ERROR. SEE APAR PK15713  FOR DESCRIPTION
  

APAR Information

• APAR is sysrouted FROM one or more of the following:

  PK09111

• APAR is sysrouted TO one or more of the following:

  UK06950

Modules/Macros

•    DESSOCK  DFHSOCK
  

Fix information

• Fixed component name

  CICSTS 3.1 Z/OS

• Fixed component ID

  5655M1500

Applicable component levels

• R400 PSY UK06950

     UP05/09/10 P F509

Fix is available

• Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.